Token emissions are one of the oldest tools in DeFi - and one of the most consistently misunderstood. The headline number is APY. The mechanism underneath is a protocol subsidising user behaviour with freshly minted tokens. When that subsidy holds, the protocol looks healthy. When it tapers, the question becomes whether the underlying product is worth using without the incentive. For most of the protocols we've tracked, the answer has been no - and the market has said so loudly.
How emissions-based incentives actually work
When a protocol launches liquidity mining rewards, it is creating demand for its token through a feedback loop: deposit capital to earn tokens, tokens trade at a premium because others want the yield, higher token price inflates the APY, higher APY attracts more capital. The loop is real and it works - for a while.
The mechanism is straightforward on-chain. A rewards contract receives a pre-determined allocation of protocol tokens per block or per epoch. Depositors earn a pro-rata share of those emissions based on their share of the pool. The APY at any given moment is a function of the emission rate, the current token price, and the total value locked. All three of those variables move.
The emission rate is typically fixed by the protocol's tokenomics schedule. Token price is market-determined. TVL is determined by depositor behaviour. The instability in the system comes from how these three interact: as emission rate falls over time (almost universally - protocols can't sustain high inflation forever), APY falls unless token price rises enough to compensate. Token price rarely rises in proportion to falling emissions. TVL therefore falls. Falling TVL per the same emission rate actually pushes APY back up temporarily, attracting new capital - until it falls again, harder.
Token emissions are a subsidy. The question is never whether the subsidy works - it always works while it's running. The question is what happens to the protocol when the subsidy stops. If the answer is "nothing much changes," the tokenomics are sound. If the answer is "most of the TVL leaves," the protocol was renting capital it never owned.
Mercenary capital and what it actually does to a protocol
The phrase "mercenary capital" is used loosely in DeFi. For the purposes of this analysis, we define it precisely: capital that enters a protocol specifically because of token incentives and whose exit decision is based primarily on incentive economics rather than belief in the protocol's underlying product.
Mercenary capital is not inherently bad. In the early days of a protocol, when organic usage is low and the product needs liquidity depth to function, subsidised capital fills a necessary gap. The problem is not that it exists. The problem is when protocol teams design around the assumption that it will stay.
On-chain behaviour distinguishes mercenary capital from sticky capital in two ways. First, wallet addresses that entered during high-emission periods show a statistically significant correlation between position exit timing and epoch transitions - when rewards drop, withdrawals cluster in the 48–72 hours following the emission reduction. Second, mercenary capital tends to arrive in large tranches from a small number of addresses (often aggregators or yield optimisers like Yearn or Beefy), and exits in the same way. The TVL chart doesn't look like slow erosion - it looks like a cliff.
The three collapsed protocols
We tracked three DeFi protocols from launch through collapse over the period January 2025 to April 2026. We are not naming them directly - two are still in wind-down proceedings with active community discussions, and one has a legal dispute pending around treasury management. The patterns, not the names, are what matters here.
Protocol A - Yield aggregator with native token rewards
A multi-chain yield aggregator that launched with a first-year emission rate of 18 million tokens per month, tapering by 30% every six months. At launch TVL, this implied roughly 280% APY on the base staking vault. Peak TVL reached $340M within the first four months. When the first emission cut triggered at month six, TVL dropped from $340M to $190M within three weeks. By month twelve, TVL was below $40M and the founding team had significantly reduced headcount.
Protocol B - Lending protocol with liquidity mining
A lending protocol offering token incentives to both borrowers and lenders, designed to bootstrap two-sided liquidity simultaneously. The emission schedule was aggressive on the lending side: 24% of total supply allocated to lender rewards in year one. The APY on stablecoin deposits reached 38% at peak, attracting significant stablecoin capital. When competitor protocols reduced their rates following a broader market correction, this protocol's APY became relatively more attractive and TVL briefly spiked - then collapsed entirely when the team announced a revised emission schedule cutting year-two rewards by 60%. The collapse took eleven days.
Protocol C - DEX with token-incentivised liquidity
A decentralised exchange that used token emissions to bootstrap trading liquidity, with rewards allocated proportionally to trading volume-weighted LP positions. The design was more sophisticated than A or B - rewards were not purely time-based but tied to productive liquidity provision. Despite this, the protocol still collapsed. The reason is instructive: when emission rates dropped, LPs didn't just reduce positions - they migrated entirely to a competing DEX that had launched with a new high-emission cycle. The trading volume followed the liquidity, not the product.
All three protocols had public tokenomics documentation. All three included language about "transitioning to sustainable fee revenue." None of them modelled what TVL retention would look like at year-two emission rates. The gap between stated plan and modelled economics is where most collapses originate.
The emissions curve they all shared
The specific numbers differed. The curve shape did not. All three protocols used what we now call a front-loaded convex decay schedule: extremely high emissions in the first 3–6 months, a steep reduction cliff at a pre-announced date, then a slower trailing tail.
The appeal of this structure is obvious from a protocol team's perspective. It maximises initial TVL growth, which drives awareness, feeds metrics, and helps justify further fundraising or partnership discussions. The cliff is announced in advance so it appears orderly. The tail is designed to look "sustainable."
The problem is in the cliff. Pre-announced cliffs are not orderly - they are coordinated exit triggers. Every participant who entered for yield knows the date. Sophisticated yield farmers set calendar reminders. Aggregator protocols update their strategy logic to begin rotation before the cliff, not after. The cliff doesn't gradually slow the flow of capital out of the protocol - it concentrates the exit into a window, which is the worst possible liquidity event for token price.
Falling token price during the exit window further depresses the APY on remaining positions, which triggers further exits. The protocol is now in a reinforcing loop it cannot escape through any mechanism available to it in real time.
Anatomy of the death spiral - the five-stage sequence
Across all three collapses, the same five-stage sequence played out with near-identical timing relative to the emission cliff. Understanding the stages separately matters because each one has a different point of intervention.
Stage 1 - Signal anticipation (weeks –4 to –1 before cliff)
Sophisticated participants begin reducing positions before the cliff. On-chain data shows gradual TVL decline in the 3–4 weeks before a scheduled emission reduction for all three protocols. The decline is not alarming at this stage - typically 8–15% off peak - but it is consistent and directional. Governance forums begin seeing threads questioning the emission schedule. Token price starts to underperform broader market benchmarks.
Stage 2 - Cliff trigger (days 0–3)
The on-chain emission rate drops as scheduled. Within 72 hours, TVL begins falling sharply. For all three protocols, the 72-hour post-cliff TVL decline was between 28% and 44%. This is the capital that had been waiting for the signal - it moves quickly because it had already decided to move and was waiting for confirmation.
Stage 3 - Token price compression (days 3–14)
Exiting participants sell their accumulated token rewards as they withdraw capital. This sell pressure hits a token whose demand curve was already dependent on the APY incentive - fewer buyers remain precisely when supply is increasing. Token price typically fell 35–55% in the two weeks following the cliff across our three case studies. The falling token price further reduces APY for remaining depositors, which accelerates Stage 4.
Stage 4 - Secondary exit wave (weeks 3–6)
Depositors who had been patient or slow to respond now face both reduced APY and a declining token price on unrealised rewards. The decision to exit becomes compelling to a much wider group. This wave is often larger in absolute TVL terms than the cliff trigger - it is the long tail of participants who were not actively monitoring positions and react to protocol forum posts, price alerts, or social media coverage of the Stage 3 decline.
Stage 5 - Liquidity death spiral (weeks 6–16)
At reduced TVL, the protocol's core product begins to degrade. For a DEX, thinner liquidity means worse execution for traders, which reduces volume, which reduces fee revenue, which makes the protocol less attractive independent of incentives. For a lending protocol, reduced deposit TVL means reduced borrowing capacity, which means reduced interest revenue, which forces further token emissions to compensate - at precisely the time when the market has demonstrated it won't absorb those emissions. The protocol is now losing on every front simultaneously.
A pre-announced emission cliff is not a managed transition. It is a coordinated exit window. Every sophisticated participant knows the date, and they act before it, not after. The cliff does not slow the outflow - it concentrates it.
— Sequere, ADMIN, SequereWhat the on-chain data shows across the three cases
| Protocol | Peak TVL | TVL at 30 days post-cliff | TVL decline | Token price decline (30d) | Survival at 6 months |
|---|---|---|---|---|---|
| Protocol A (yield aggregator) | $340M | $52M | –85% | –71% | Reduced ops, no new dev |
| Protocol B (lending protocol) | $218M | $19M | –91% | –83% | Wind-down in progress |
| Protocol C (DEX) | $127M | $28M | –78% | –54% | Pivot to different product |
| Uniswap v3 (baseline) | — | Stable | <5% over same period | –12% (market-correlated) | Growing |
The Uniswap v3 comparison is important context. Uniswap does not have ongoing liquidity mining rewards at scale. Its TVL is sticky because liquidity providers earn real trading fees from real volume. During the same period when all three case study protocols were collapsing, Uniswap v3 TVL declined by less than 5% - a decline correlated with broader ETH price movement, not incentive structure.
This is the fundamental test for any emissions-based tokenomics design: what percentage of your TVL stays when the emissions stop? If you cannot answer that question with data, you do not understand your protocol's real product-market fit.
The three tokenomics design mistakes driving every collapse
Behind the emissions curve, three specific design errors recur. They are addressable. The protocols that avoid them consistently perform better over multi-year windows.
Mistake 1 - Treating APY as the product
When a protocol's growth story is primarily about APY numbers, the capital it attracts is primarily responding to APY numbers. This sounds obvious but is routinely ignored in launch planning. Protocols that market themselves through yield comparisons - especially those that top yield aggregator leaderboards - are selecting for exactly the capital that will leave when the yield comparison turns unfavourable.
The alternative is designing emissions to reward behaviours that build the protocol's real moat: long-term liquidity provision, governance participation, referrals, integrations. Rewards tied to these behaviours attract capital and participants with longer time horizons. They are also harder to game at scale, which reduces mercenary capture.
Mistake 2 - Front-loading to maximise early metrics
The incentive for a protocol team to front-load emissions is real and understandable: high early TVL enables fundraising, generates press coverage, and gives the protocol a strong negotiating position for integrations and listings. The problem is that this strategy borrows from the future. Every token emitted in month one at high APY is a unit of sell pressure that will be realised at some point, and a signal about protocol value that will have to be maintained or unwound.
Protocols that use a more graduated emission schedule - even one that delivers lower initial APY - tend to attract a higher ratio of sticky to mercenary capital at each stage, because the yield gap between their protocol and competitors is smaller and therefore less worth farming specifically for the incentive. Lower headline APY can produce better TVL retention over 12 months.
Mistake 3 - No treasury backstop for the transition period
The period between high emissions and genuine fee revenue generation is the most dangerous in a protocol's lifecycle. Protocols that survive it typically have one thing in common: treasury reserves sufficient to supplement emissions during the transition, buying time for organic usage to grow. Of the three collapsed protocols in our study, none had a treasury backstop of more than 90 days of operating costs at their peak TVL operating level. All three ran out of runway to cover the gap between "subsidised usage" and "earned usage."
Emission structures that hold up - what the survivors do differently
The protocols that have maintained TVL and token price over 18+ month windows share a different structural approach to emissions. None of them are frictionless - they involve design trade-offs that feel expensive at launch but pay off over time.
Graduated, low-cliff schedules
Instead of a single large reduction at a fixed date, emissions decline gradually - monthly or quarterly by small percentages. This removes the coordinated exit trigger. There is no single date that signals "get out now." Sophisticated farmers can still model the declining return curve, but the exit is distributed over many months rather than concentrated into days. Real-world examples of protocols using this approach show TVL declining at roughly the same rate as emissions rather than faster, which is the characteristic signature of a sticky-capital base.
Fee sharing that scales with usage
Protocols that replace declining token emissions with genuine fee revenue sharing create a pull incentive that grows as the protocol's usage grows. The mechanics vary - some use buybacks, some distribute ETH or stablecoins directly, some accrue value to staked governance tokens - but the underlying dynamic is the same: depositors who stay beyond the high-emission period are earning returns tied to real economic activity, not inflation. This is the only mechanism we have seen reliably retain capital through emission reductions.
Emission velocity limits on large positions
Some protocols have implemented per-address or per-epoch withdrawal velocity limits on reward accrual. The intent is to reduce the structural advantage of fast-exit infrastructure that sophisticated farmers operate. In practice, the effectiveness is mixed - sufficiently sophisticated participants use multiple addresses - but the friction cost is real and does meaningfully reduce the speed of the Stage 2 and Stage 4 exit waves described above.
No emissions schedule survives contact with a protocol that does not have genuine product-market fit. The healthiest emission designs buy time for the product to find its natural user base. If the product cannot find that base within the emission runway, the schedule is irrelevant - the collapse happens either with or without a well-designed curve.
Design checklist for protocol builders
The following questions are the ones we now ask before advising on any tokenomics design that includes an emissions component. They are not exhaustive, but every protocol that skipped more than two of them ended up in the failure pattern described above.
Before launch:
- What is your TVL retention assumption at year-two emission rates? Model it explicitly. If you cannot estimate it, you have not thought through your user segmentation.
- What percentage of your current TVL exists because of the incentive, and what percentage because of the product? Use on-chain cohort analysis: which depositor addresses also use the protocol's core function, and which only interact with the reward contract?
- How long can your treasury cover operations at half of peak TVL? The answer should be at least 12 months. If it is less, you are structurally dependent on the emission cycle working as planned, which is the highest-risk assumption in the entire model.
- Is there a cliff in your emission schedule? If yes, remove it or convert it to a gradual slope. Pre-announced cliffs are exit triggers, not managed transitions.
After launch:
- Monitor the ratio of emission-only participants to product users weekly. A rising ratio is an early warning that the incentive is detaching from genuine product demand - usually 3–4 months before it shows up in TVL or token price.
- Do not announce emission reductions in governance votes without a simultaneous announcement of what replaces the yield. Governance forums are public. Every participant reads them. A reduction vote without a replacement narrative is a liquidation signal.
- Build fee-sharing mechanisms before you need them. Protocols that try to implement buybacks or fee distribution after a TVL decline are doing it in a low-liquidity environment where it is most costly and least effective. The infrastructure needs to be live while TVL is high.
The death spiral is not inevitable. It is the predictable result of a specific set of design choices that have been made often enough that we now have a clear failure pattern to map against. Protocols that treat token emissions as a permanent product feature rather than a temporary bootstrapping tool, that design around sticky capital rather than mercenary capital, and that build genuine fee revenue before the emission runway ends - those protocols do not show up in case studies like this one.
If you are designing tokenomics for a new protocol or auditing an existing emission schedule, we are glad to work through the specifics. Reach out here or book a tokenomics review session with the team.